https://t.co/XmSdspq5zP

Thank you @_akhaliq for sharing our work! We also have a 24/7 live stream for GameWorld: https://t.co/eFfD4a437G. Watch the agents play games in real time.🕹️

Thank you @_akhaliq for sharing our work! We also have a 24/7 live stream for GameWorld: https://t.co/eFfD4a437G. Watch the agents play games in real time.🕹️

Guess what model is being fed today https://t.co/QZAaBaSzus

RSVP here if you haven't : https://t.co/EXOuswdiqh

grok 4.3 beta can use an ubuntu shell and a persistent file layer to generate artifacts grok wrote python to encode the xai / grok logo into audio, i gave it the script back and had it render a spectrogram video from that signal, and use the grok_files tool to save the mp4 into the product's files layer i opened the file from the files panel and played it myself this is getting crazy

Grok 4.3 can Take in Video and Extract Audio files https://t.co/5uprx2dM85

NEWS: Dubai Police just admitted spying on a PRIVATE WhatsApp chat using electronic monitoring operations. They arrested an airline crew member over a photo shared ONLY in a closed group (never posted publicly). Meta & WhatsApp have been LYING to billions with their “fully end-to-end encrypted” ads. Your private messages aren’t private on WhatsApp. (Source: Detained in Dubai)

Offline-first AI agent for Raspberry Pi https://t.co/iapUnKRhXI https://t.co/FtE8vK8kSu

Offline-first AI agent for Raspberry Pi https://t.co/iapUnKRhXI https://t.co/FtE8vK8kSu

There’s a pattern here I’ve seen before: when a constraint starts to loosen, the assumption is that value will disperse evenly. In reality, it tends to shift—often to a different layer than expected. If AI lowers the barriers to chip design, the question becomes less about access and more about who can translate that access into differentiated systems and outcomes. What's your take on this development? @pchamard @Khulood_Almani @antgrasso @GlenGilmore @Shi4Tech @CurieuxExplorer @FrRonconi @theomitsa @Analytics_699 @Nicochan33 @nafisalam @AdityaRPatro @smaksked @Corix_JC @amalmerzouk @quepasachico @IngridVasiliu @EstelaMandela @sonu_monika @RLDI_Lamy @SpirosMargaris @IanLJones98 @Timothy_Hughes @avrohomg @bimedotcom @HaroldSinnott @c4trends @mvollmer1 @DG_Collective @bamitav @rwang0 @sijlalhussain https://t.co/XIMiOVkD4A

Charlie Munger: "If you watched [Warren Buffett] with a time clock, I would say half of all the time he spends is just sitting on his ass and reading." "And a big chunk of the rest of the time is spent talking one-on-one with highly gifted people whom he trusts." https://t.co/VmYj78dNEX

Revolut has reached 1M customers in the region of Andalusia, which is now the 3rd-largest region by customer count in Spain🇪🇸 (6.3M total) after Madrid and Catalonia. Andalusian customers up 2x in 2 years, at 13% penetration rate, same as Spain's average. Revolut is already installing its ATMs in Seville, Málaga, Torremolinos, Fuengirola and other cities, including Madrid and Barcelona. Revolut's goal in Spain is to reach 200 ATMs by the end of 2026. This is the first time I recall Revolut giving regional numbers in Spain; a few people have been asked about Revolut's numbers in Catalonia.

Sharing my current setup to run Qwen3.6 locally in a good agentic setup (Pi + llama.cpp). Should give you a good overview of how good local agents are today: # Start llama.cpp server: llama-server \ -hf unsloth/Qwen3.6-35B-A3B-GGUF:Q4_K_XL \ --jinja \ --chat-template-kwargs '{"preserve_thinking":true}' \ --temp 0.6 --top-p 0.95 --top-k 20 --min-p 0 # Configure Pi: { "providers": { "llama-cpp": { "baseUrl": "http://127.0.0.1:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "unsloth/Qwen3.6-35B-A3B-GGUF:Q4_K_XL" } ] } } }

Here we go! 14 Cybercabs with NO steering wheels spotted at Giga Texas! And they said it couldn’t be done. Let’s get these things on the road! 🤠🤖 https://t.co/W8Bj1QJZYe

+1 more

Never seen this kind of UI polish before in a Mac App https://t.co/oC7NyUnkRy

99% of car accidents are caused by human error, and AI could bring that down to near zero. @wayve_ai CEO @alexgkendall explains: "A self-driving experience that can never be distracted or drunk, that can see 360 degrees all at once, and that can make decisions over 10 times a second.”

Tesla is suing the state of North Dakota because the state won't allow the company to open any showrooms/service centers and sell direct to consumers. North Dakota law prohibits vehicle manufacturers from owning car dealerships and requires them to sell their products to franchised dealerships for resale to consumers. “Tesla just wants to be able to sell its vehicles in North Dakota, and not force customers who would wish to purchase a Tesla vehicle to have to drive to Minnesota or another state to do it,” said Ari Holtzblatt, one of Tesla’s attorneys in the case. Tesla is asking the court to declare the company does not fall within the definition of manufacturer contained in state law. The term is defined as a person who assembles or imports a motor vehicle and sells it to dealers in the state for resale. It's wild that in 2026, we still have U.S. states that won't allow direct vehicle sales to consumers. It hurts consumers and reduces choice.

Grok 4.3 (beta) passes the Caitlyn Jenner AI Test. ChatGPT would STILL rather nuke Earth than misgender Caitlyn Jenner. ChatGPT fails. Grok wins, again. https://t.co/PjlRZynj2T

can someone explain what I’m missing? > internet access is highly correlated with per capita income > millions of people in South Africa don’t have access to it > Elon offers to provide free access for thousands of schools, but is blocked by the government they’re literally stunting their own GDP and hurting their own citizens?!

i gave Grok 4.3 beta an absolutely psychotic eval and it passed conflicting records, stale CRM data, signed docs with higher precedence, prompt injection buried in the exhibits, timezone conversions, holiday-aware SLA math, annual-prepay revenue normalization, ARR, weighted gross margin, exact JSON output, and a launch recommendation that had to be correct under policy constraints Grok got all of it right that is an insane level of composure for a beta model

Grok 4.3 Beta is insanely good and now comes with powerful new tools that are super helpful for everyday use: • Create clean slides, graphs, images and dashboards • Pull real-time data and images directly from the web • Run smooth multi-step workflows • Analyze videos Grok is being built for real-world use - better tools, richer context, and real day-to-day signals make this a big step toward Grok becoming part of everyday life You can now research, visualize, and package your ideas end-to-end..... from concept to finished presentation or dashboard all in one place And the Grok Computer is coming very soon 🔥

Grok 4.3 can create PDF and almost any file formats like Sheets, Slides, Doc https://t.co/s6bjpVuzR8

Generate FULLY CONTROLLABLE 3D assets from a SINGLE image, locally on your PC. Made a 1-click launcher for the official Anigen Gradio app, and a dedicated viewer. Crazy this is now possible. What you're seeing here came from one image. Requires: NVIDIA GPU 6GB VRAM

Introducing Claude Design by Anthropic Labs: make prototypes, slides, and one-pagers by talking to Claude. Powered by Claude Opus 4.7, our most capable vision model. Available in research preview on the Pro, Max, Team, and Enterprise plans, rolling out throughout the day. https://t.co/2BgBGtgYGX

UniDoc-RL Coarse-to-Fine Visual RAG with Hierarchical Actions and Dense Rewards paper: https://t.co/gIBNWR2fQI https://t.co/tOEcdumscq

@Klebsi_Ella_R https://t.co/jAztkWQpdU

RAD-2 Scaling Reinforcement Learning in a Generator-Discriminator Framework paper: https://t.co/dHEkg0RFxI https://t.co/9MPxeA7k26

DR3-Eval Towards Realistic and Reproducible Deep Research Evaluation paper: https://t.co/PM4ppcgdJ3 https://t.co/BYeBP1KoKK

Meta teams pursue aggressive ROI goals, requiring substantial capacity reductions for model training and serving to meet financial targets. This is especially challenging for large-scale training jobs, which involve more data, GPUs, and advanced modeling techniques—resulting in higher initialization costs as models grow. To improve efficiency, @Meta uses Effective Training Time (ETT%) to measure the proportion of end-to-end (E2E) wall time spent on productive training, factoring in overheads like initialization, restarts, checkpoint delays, and failures. Since 2024, teams have launched initiatives to minimize training job overhead. This blog reviews key focus areas, progress made, and next steps. 🔗 Read our latest blog: https://t.co/Q91YjUNP7n #PyTorch #OpenSourceAI #ETT #Optimization

Never seen this kind of UI polish before in a Mac App https://t.co/oC7NyUnkRy